The Side-Effect Effect and Our Concept of

Intentional Action

If pragmatic considerations don't adequately explain the side-effect effect, Knobe (2003a, 2006) argues that maybe the effect is telling us something very specific about the relationship between normative judgments and people's concept of intentional action and about the proper function of the underlying mechanism responsible for people's judgments about intentional action.

Thomas Nadelhoffer (2004a, 2004b, 2006b) agrees with Knobe that these results tell us something about the relationship between people's moral judgments and their concept of intentional action. However, he denies that these results tell us something about the proper function of the underlying mechanism responsible for people's judgments about intentional action, arguing instead that they reveal ways in which this mechanism might malfunction.¹⁰ Taking Knobe's Environmental Harm case as our model, the general idea is this: people who are asked to consider the Environmental Harm case have a negative affective response to the chairman; this negative affective response distorts people's intentionality judgments, causing them to incorrectly judge that the chairman harmed the environment intentionally.¹¹

Claiming that people's negative affective judgments are distorting their intentionality judgments, of course, suggests that moral considerations shouldn’t affect our intentionality judgments.

Police Officer:

Imagine that a thief is driving a car full of recently stolen goods. While he is waiting at a red light, a police officer comes up to the window of the car while brandishing a gun. When he sees the officer, the thief speeds off through the intersection. Amazingly, the officer manages to hold on to the side of the car as it speeds off. The thief swerves in a zigzag fashion in the hope of escaping - knowing full well that doing so places the officer in grave danger. But the thief doesn't care; he just wants to get away. Unfortunately for the officer, the thief's attempt to shake him off is successful. As a result, the officer rolls into oncoming traffic and sustains fatal injuries. He dies minutes later.

Car Thief:

Imagine that a man is waiting in his car at a red light. Suddenly, a car thief approaches his window while brandishing a gun. When he sees the thief, the driver panics and speeds off through the intersection. Amazingly, the thief manages to hold on to the side of the car as it speeds off. The driver swerves in a zigzag fashion in the hope of escaping - knowing full well that doing so places the thief in grave danger. But the driver doesn't care; he just wants to get away. Unfortunately for the thief, the driver's attempt to shake him off is successful. As a result, the thief rolls into oncoming traffic and sustains fatal injuries.

Although the two vignettes are structurally identical, Nadelhoffer (2006b) found that people who are asked to consider the Police Officer case form a stronger negative affective response towards the driver of the car than people who are asked to consider the Car Thief case.¹² Additionally, people who are asked to consider the Police Officer case are more likely to judge that the driver caused the death intentionally than people who are asked to consider the Car Thief case.¹³ This means that people are more likely to hold the driver legally responsible for the death of the police officer than the death of the car thief. But, according to Nadelhoffer, we don't want the fact that someone is charged with a crime that generates strong negative affective responses to make it more likely that people will find her legally responsible for the crime in question. As such, Nadelhoffer urges that moral considerations shouldn’t affect our intentionality judgments. The influence of moral considerations on our intentionality judgments signals a malfunction of the underlying mechanism responsible for our judgments about intentional action.

Let's look more closely at this suggestion. The central idea is that people's affective responses are driving (in this case, somewhat recklessly) their intentionality judgments. People feel that the chairman deserves blame for harming the environment and this negative affective response causes them to (incorrectly) judge that he did so intentionally. The problem is that intentionality judgments aren't always emotional. In fact, we find that even people (for example, patients with ventromedial prefrontal cortex damage) who lack the ability to have any affective response to the chairman's actions nevertheless judge that the chairman harmed the environment intentionally (Young et al. 2006). This calls into question the suggestion that it is an emotional response to the chairman's actions that is causing people to judge that he harmed the environmental intentionally.

Joshua Knobe and Thomas Nadelhoffer agree that these results tell us something about the relationship between people's moral judgments and their concept of intentional action. Knobe argues these results tell us something about the proper function of the underlying mechanism responsible for people's judgments about intentional action, while Nadelhoffer argues that these results tell us something about how this mechanism might malfunction. Edouard Machery (2008) challenges both of these views. First, Machery claims that we should remain neutral about whether or not the side­effect effect is best explained in terms of competent intentionality judgments. Second, he claims that the results actually tell us something, not about the relationship between people's moral judgments and their concept of intentional action, but about the relationship between people's concept of intentional action and considerations of costs and benefits.

face=Arial>Before turning our attention to the second claim, which has garnered the most attention, it is worth spending a few moments looking at Machery's call for neutrality.

Nadelhoffer avoids this problem by simply stipulating what kind of work our folk concept of intentional action is supposed to be doing for us. This allows him to maintain that the side-effect effect tells us something about people's conceptual performance, but does so in a way that leaves open the possibility that someone else (Knobe, for example) might simply stipulate a different account according to which the side-effect effect tells us something about people's conceptual competence.

With that discussion out of the way, let's look more closely at Machery's second claim, namely that the results actually tell us something, not about the relationship between people's moral judgments and their concept of intentional action, but about the relationship between people's concept of intentional action and considerations of costs and benefits. Recall that Knobe (2003a) found that most people asked to consider the Environmental Harm case judge that the chairman intentionally harmed the environment, while most people asked to consider the Environmental Help case judge that the chairman did not intentionally help the environment. To explain these results, Machery proposes what he calls the trade-off hypothesis. There is a cost associated with increasing profits in the Environmental Harm case, namely harming the environment. Since people believe that costs are incurred intentionally, this explains why people asked to consider the Environmental Harm case judge that the chairman intentionally harmed the environment. By contrast, there are no costs associated with increasing profits in the Environmental Help case. This explains why people asked to consider the Environmental Help case judge that the chairman did not intentionally help the environment.

Machery supports the trade-off hypothesis with two pairs of vignettes. Let's begin with the first pair:

Extra Dollar:

Joe was feeling quite dehydrated, so he stopped by the local smoothie shop to buy the largest sized drink available. Before ordering, the cashier told him that the Mega-Sized Smoothies were now one dollar more than they used to be. Joe replied, “I don't care if I have to pay one dollar more, I just want the biggest smoothie you have.” Sure enough, Joe received the Mega-Sized Smoothie and paid one dollar more for it.¹⁸

Free Cup:

Joe was feeling quite dehydrated, so he stopped by the local smoothie shop to buy the largest sized drink available. Before ordering, the cashier told him that if he bought a Mega-Sized Smoothie he would get it in a special commemorative cup. Joe replied, “I don't care about a commemorative cup, I just want the biggest smoothie you have.” Sure enough, Joe received the Mega­Sized Smoothie in a commemorative cup.

Machery found that most people asked to consider the Extra Dollar case judge that Joe intentionally paid an extra dollar, while most people asked to consider the Free Cup case judge that Joe did not intentionally buy the commemorative cup.¹⁹ Since most people viewed the actions (paying an extra dollar and buy a commemorative cup) as morally neutral, this provides evidence that something other than people's beliefs about the moral status of an action (or the outcome of that action) influence their intuitions about whether or not the action was performed intentionally. Instead, this is evidence that people's intuitions about whether or not an action was performed intentionally are being influenced by simple considerations of costs and benefits.

The apparent influence of simple considerations of costs and benefits on people's intentionality judgments is not limited to cases involving morally neutral actions. The same influence appears to be found, for example, in cases involving morally appropriate actions. To see this, let's consider Machery's second pair of vignettes:

Worker:

John is standing near the tracks of a trolley. John notices that the brakes of the trolley have failed. Five workmen are working on the tracks with their backs turned. John sees that the runaway trolley is headed for the five workmen who will be killed if it proceeds on its present course. The only way to save these five workmen is to hit a switch that will turn the trolley onto the side tracks. Unfortunately, there is a single workman on the side tracks with his back turned. John knows that the workman on the side tracks will be killed if he hits the switch, but the five workmen will be saved. John decides to hit the switch. Sure enough, the trolley turns on the sidetracks, the five workmen on the main tracks are saved, and the workman on the side tracks is killed.

Dog:

John is standing near the tracks of a trolley. John notices that the brakes of the trolley have failed. Five workmen are working on the tracks with their backs turned. John sees that the runaway trolley is headed for the five workmen who will be killed if it proceeds on its present course. The only way to save these five workmen is to hit a switch that will turn the trolley onto the side tracks. Moreover, there is a dog on the tracks with its back turned. John knows that the five workmen and the dog will be saved if he hits the switch. John thinks “I don't care at all about saving the dog. I just want to save the five workmen.” John decides to hit the switch. Sure enough, the trolley turns on the side tracks, the five workmen and the dog on the main tracks are saved.

Machery found that most people asked to consider the Worker case judge that John intentionally caused the death of the workman on the side tracks, while most people asked to consider the Dog case judge that John did not intentionally save the dog.²⁰ Since most people viewed the actions (causing the death of the workman on the side tracks and saving the dog) as morally appropriate actions, this provides additional evidence that something other than people's beliefs about the moral status of an action (or the outcome of that action) influence their intuitions about whether or not the action was performed intentionally.²¹ Again, it appears that people's intuitions about whether or not an action was performed intentionally are being influenced by simple considerations of costs and benefits.

Let's look more closely at Machery's trade-off hypothesis, according to which people's intuitions about whether or not an action was performed intentionally are being influenced by simple considerations of costs and benefits. There are two possible readings of this hypothesis.²² On one reading of the hypothesis, people's intuitions about whether or not an action was performed intentionally are being influenced by the fictional agent's perception of something as a cost. Let's call this reading, the agent-oriented reading. On a second reading of the hypothesis, people's intuitions about whether or not an action was performed intentionally are being influenced by their own perceptions of something as a cost. Let's call this reading, the subject-oriented reading.

The trouble is that there seem to be significant problems for the trade-off hypothesis on either reading. Let's start with the agent- oriented reading. Ronald Mallon (2009) points out that if people's intuitions about whether or not an action was performed intentionally are being influenced by the fictional agent's perception of something as a cost, then we should expect the side-effect effect to disappear in cases where the fictional agents don't view the relevant side effects as costs. But consider the following two vignettes:

Harmful Terrorist:

A member of a terrorist cell went to the leader and said, “We are thinking of bombing a nightclub. It will kill many Americans, but it will also harm the Australians since many Australians will be killed too.” The leader answered, “I admit that it would be good to harm the Australians, but I don't really care about that. I just want to kill as many Americans as possible! Let's bomb the nightclub!” They did bomb the nightclub, and sure enough, the Australians were harmed since many Australians were killed.

Helpful Terrorist:

A member of a terrorist cell went to the leader and said, “We are thinking of bombing a nightclub. It will kill many Americans, but it will also drive down property costs, helping the nearby orphanage acquire the land it needs for the children.” The leader answered, “I admit that it would be good to help the orphanage, but I don't really care about that. I just want to kill as many Americans as possible! Let's bomb the nightclub!” They did bomb the nightclub, and sure enough, the orphanage was helped by falling property values.

Mallon (2009) found that most people asked to consider the Harmful Terrorist case judge that the terrorist intentionally harmed the Australians, while most people asked to consider the Helpful Terrorist case judge that the terrorist did not intentionally help the orphanage.²³ This provides evidence against the view that people's intuitions about whether or not an action was performed intentionally are being influenced by the fictional agent's perception of something as a cost. Something else must be influencing people's intuitions about whether or not the action was performed intentionally.

Maybe the subject-oriented reading is right. Mark Phelan and Hagop Sarkissian (2009) point out that if people's intuitions about whether or not an action was performed intentionally are being influenced by their own perception of something as a cost, then we should expect that most people are willing to judge that an action is intentional when they view it as a cost to be incurred in the pursuit of some important goal. But consider the following vignette:

Caring Lieutenant:

A lieutenant was talking with a sergeant. The lieutenant gave the order: “Send your squad to the top of Thompson Hill.” The sergeant said: “But if I send my squad to the top of Thompson Hill, we'll be moving the men directly into the enemy's line of fire. Some of them will surely be killed!” The lieutenant answered: “Look, I know that they'll be in the line of fire, and I know that some of them will be killed. I care about my soldiers more than anyone else. But it's imperative to the success of this campaign that we take Thompson Hill.” The squad was sent to the top of Thompson Hill. As expected, the soldiers were moved into the enemy's line of fire, and some of them were killed.

Phelan and Sarkissian found that most people asked to consider the Caring Lieutenant case judge that the lieutenant did not intentionally cause the soldiers' deaths. Since people generally perceive the soldiers' deaths as a cost, this provides evidence against the view that people's intuitions about whether or not an action was performed intentionally are being influenced by their own perceptions of something as a cost.²⁴·²⁵ Again, something else must be influencing people's intuitions about whether not the action was performed intentionally.

So far we've been considering explanations of the side-effect effect according to which these results tell us something specific about the relationship between normative judgments and people's concept of intentional action. All of these explanations assume that there is a single folk concept of intentional action. Shaun Nichols and Joseph Ulatowski (2007) challenge this assumption, arguing both that there are multiple folk concepts of intentional action and that the side-effect effect is better explained as telling us something about the relationship between normative judgments and these different folk concepts of intentional action.

At issue is how best to explain minority responses. While most people asked to consider the Environmental Harm case judge that the chairman intentionally harmed the environment and most people asked to consider the Environmental Help case judge that the chairman did not intentionally help the environment, responses aren't univocal. Some people asked to consider the Environmental Harm case judge that the chairman did not intentionally harm the environment and some people asked to consider the Environmental Help case judge that the chairman intentionally helped the environment. How should we explain these minority responses?

Typically, such responses are treated as noise resulting from some kind of performance error. Nichols and Ulatowski (2007) suggest a different explanation. They asked people to consider both the Environmental Harm case and the Environmental Help case. They found an interesting pattern of responses. Some people asked to consider both cases provide asymmetric responses, judging that the chairman intentionally harmed the environment but did not intentionally help the environment. Other people asked to consider both cases provide symmetric responses, judging either that the chairman both intentionally harmed the environment and intentionally helped the environment or that the chairman neither intentionally harmed the environment nor intentionally helped the environment. Additionally, they found that people who judge that the chairman intentionally harmed the environment typically explain their judgments by appealing to the chairman's belief that the environment would be harmed and that people who judge that the chairman did not intentionally help the environment typically explain their judgments by appealing to the chairman's lacking the desire to help the environment. This pattern of responses, and corresponding set of explanations, leads Nichols and Ulatowski to suggest that there are actually two separate folk concepts of intentional action, one involving an epistemic component and the other involving a motivational component.

Fiery Cushman and Alfred Mele (2008) point out a problem with Nichols and Ulatowski's view that there are two distinct concepts of intentional action, one involving an epistemic component and the other involving a motivational component. In both the Environmental Harm case and the Environmental Help case, the chairman has a belief about what impact the profit-making venture will have on the environment but lacks the relevant desire to either harm or help the environment. Cushman and Mele wonder how people would respond to cases in which the agent has the relevant desire but lacks the relevant belief. Cushman and Mele predict that, if Nichols and Ulatowski are right, three things should happen. First, people working with the epistemic concept should be more likely to judge that an agent has acted intentionally when she has the relevant belief but lacks the relevant desire than when she has the relevant desire but lacks the relevant belief. Second, people working with the motivational concept should be more likely to judge that an agent has acted intentionally when she has the relevant desire but lacks the relevant belief than when she has the relevant belief but lacks the relevant desire. Third, people working with the motivational concept should be more likely than people working with the epistemic concept to judge that an agent has acted intentionally when she has the relevant desire but lacks the relevant belief. Contrary to two of these predictions, Cushman and Mele found that people working with the motivational concept were not more likely than people working with the epistemic concept to judge that an agent has acted intentionally when she has the relevant desire but lacks the relevant belief, and that people working with the epistemic concept were not more likely to judge that an agent has acted intentionally when she has the relevant belief but lacks the relevant desire than when she has the relevant desire but lacks the relevant belief. Cushman and Mele conclude that there are actually more than two distinct concepts of intentional action: one that treats desire as a necessary condition for intentional action; another that treats it as a sufficient condition for intentional action; and yet another that treats it as a sufficient condition for intentional action only in cases of morally bad actions.

However many concepts are involved, the advantage of a multiple concept explanation is that it explains minority responses without having to explain them away. It also provides a neat explanation of the side-effect effect itself. In the Environmental Harm case, the chairman knowingly harms the environment but does not care at all about harming it; in the Environmental Help case, the chairman knowingly helps the environment but does not care at all about helping it. What explains the fact that most people asked to consider the Environmental Harm case judge that the chairman intentionally harmed the environment, while most people asked to consider the Environmental Help case judge that the chairman did not intentionally help the environment is that different normative judgments trigger different concepts of intentional action.

5.   Conclusion

It seems that the side-effect effect is telling us something about the relationship between normative considerations and our concept - or concepts - of intentional action. Dean Pettit and Joshua Knobe (2009) suggest that the sphere of influence is even wider, arguing that it tells us something about the relationship between normative considerations and folk psychology in general. They found that normative considerations influence a wide variety of folk psychological judgments, including judgments about decision, desire, preference, advocacy, and choice. Other researchers have recently added studies that suggest that normative considerations influence folk psychological judgments about causation (Hitchcock & Knobe 2009, Roxborough & Cumby 2009) and knowledge (Beebe & Buckwalter 2010). With more studies emerging all of the time, it now seems like we are only beginning to comprehend just how widespread this influence may be.

It also remains to be seen just what the side-effect effect is actually telling us. In particular, it remains to be seen whether it is telling us something about our conceptual competence or our conceptual performance. And, here, experimental philosophers face a problem. Determining whether or not the side-effect effect is telling us something about our conceptual competence requires experimental and theoretical resources that experimental philosophers don't currently have. What is needed is some account of the kind of work our folk concepts are supposed to be doing and/or an account of the physical implementation of the relevant cognitive processes and mechanisms. Unfortunately, the kind of experimental methods typically employed by experimental philosophers, namely survey methods, aren't apt to produce either kind of account. What is needed instead, for example, are neuroanatomical accounts of the cognitive processes and mechanisms responsible for our folk psychological judgments and evolutionary (or other teleological) accounts of the work that our folk concepts are supposed to be doing. Developing these kinds of accounts is going to require experimental philosophers to become even more experimental by expanding the variety of experimental methods that they employ.